@FinanceYF5: Three Waves of AI Token Demand 1/ Over the next 4 years, AI token consumption is projected not just to double, but to grow 24-fold.

X AI KOLs Timeline News
ai-token-demand future-ai-trends ai-agents robotaxi ai-infrastructure

Summary

This article discusses three waves of AI token demand, projecting a 24-fold increase in AI token consumption over the next four years. The primary drivers are AI agents, robotics, and Robotaxis, suggesting that AI infrastructure may have just entered its second stage.

Three Waves of AI Token Demand 1/ Over the next 4 years, AI token consumption is projected not just to double, but to grow 24-fold. What will truly drive demand isn't chat, but the second wave of Agents: continuously running, batch-deployed, and constantly calling models. Further out are robotics and Robotaxis. If this chart holds, AI infrastructure might have just entered its second stage.👇 https://t.co/odBgJbTVbb
Original Article
View Cached Full Text

Cached at: 08/18/26, 08:24 AM

The Three Waves of AI Token Demand

1/ Over the next four years, AI’s token consumption is expected not to double, but to increase 24-fold.

The true driver of this demand won’t be chat, but the second wave of Agents: continuously running, deployed in batches, and constantly calling models. Further out, there are robots and Robotaxi.

If this chart holds true, AI infrastructure might now be just entering its second phase. 👇 https://t.co/odBgJbTVbb

2/ The first wave is Chat.

A human asks a question, the model answers once. It allowed hundreds of millions of people to form the habit of using AI, but individual tasks are short, invocation counts are limited, and the growth potential is gradually slowing.

Chat is just the starting point for AI demand, not the largest token consumption scenario.

3/ The second wave is Agents.

They don’t stop after one answer. Instead, they continuously plan, call tools, and check results until the task is complete.

Goldman Sachs predicts that by 2030, consumer and enterprise agents will consume 120 quadrillion tokens per month, which is 24 times the 2026 level.

4/ Enterprises might deploy hundreds or even thousands of agents simultaneously, having them handle code, customer service, finances, and internal processes around the clock.

Consumer-facing agents haven’t even been deployed at scale yet: new versions of Siri, Gemini on Android, and agents that can truly operate a phone on behalf of users could continue to amplify demand.

This is also why the demand for computing power, memory, and electricity will keep growing.

5/ The third wave is Physical AI: humanoid robots, Robotaxi, and various autonomous machines.

They run models and consume tokens around the clock while themselves requiring chips, memory, sensors, and power systems.

A single robot simultaneously drives two demand streams: cloud inference and physical hardware.

6/ However, it’s important to note that the agent data before 2030 in the chart is a Goldman Sachs forecast; the dashed lines for Physical AI in 2040 are labeled as a [Potential Upper Limit], not the baseline scenario.

What’s truly worth noting isn’t any specific number, but that AI is shifting from “occasional chat” to “continuous work,” and then into the real world.

Which wave are you most optimistic about?

Latest: Popular SaaS Providers on the Ramp Platform in July 2026

Two major trends emerged this month:

  1. From payment data, AI customer service bots have become one of the clearest enterprise AI application scenarios (PolyAI, Sierra)
  2. The usage of open-source models and Chinese models continues to grow (DeepSeek, Fireworks AI, Baseten, Together, OpenRouter)

Similar Articles

@FinanceYF5: Chamath said he asked the CTO about the company's AI token spending today and got a shocking answer: token costs double every 45 days, but downstream productivity improves by at most 5%. His exact words were that costs double, benefits stay roughly flat, and the team found that to iterate to the next generation of capabilities, the amount of tokens required grows exponentially, because...

X AI KOLs Timeline

Chamath revealed that his company found AI token costs double every 45 days while downstream productivity improves by at most 5%. He believes AI development is approaching a bottleneck and suggests companies reconsider their strategies or even consider exiting.

@DashHuang: Imagine AI as digital employees. It's normal for the salary of subordinates managed by one person to be at a 1:1 ratio, or even 1:10, with their own salary. Therefore, it is foreseeable that the cost of AI tokens will match or exceed employee salaries in the future. Of course, the measurement of ROI will be the same as for team size and salary costs.…

X AI KOLs Following

Compares AI token consumption to digital employee salaries, predicts token costs will match or exceed employee wages, and discusses how businesses measure ROI and control costs.